Evaluation and Attribution of a Warm Winter Bias over Arctic Sea Ice in a Climate Model

Abstract Winter near-surface temperature biases over Arctic sea ice in climate models contribute to uncertainties in future sea ice and Arctic climate projections. Mitigating these biases requires proper evaluation and understanding of their origin. We focus on the 1980–2014 near-surface air temperature simulated in the atmosphere-only and coupled configurations of the IPSL-CM6A-LR climate model. Model outputs are evaluated against multiple observational datasets, including the ERA5, in situ measurements, and satellite-based remote sensing products. To account for the known warm bias in ERA5 near-surface air temperatures during Arctic winters, we applied a simple linear correction fitted on in situ observations over the entire Arctic perennial ice zone, which served as the reference data. Within all observational datasets, a winter warm bias over sea ice is evident in both model configurations: a slight +0.4°C in the atmosphere-only IPSL-CM6A-LR and +3.1°C in the fully coupled configuration. A pan-Arctic energy budget analysis attributes this warm bias primarily to excess poleward atmospheric heat transport. In the coupled model, the bias is further amplified by an underestimated summertime sea ice cover, despite a deficit in net oceanic heat transport. The methods developed here could be used in multimodel evaluations to progress understanding and reduce Arctic biases in climate models. Significance Statement In recent decades, Arctic warming has been nearly 4 times faster than the global average, a phenomenon known as Arctic amplification. This rapid warming is closely linked to the melting of sea ice and land ice, with large implications for the atmosphere, the ocean, ecosystems, and human activities. Reliable climate model projections are essential for understanding and anticipating these changes. Yet, the sparse long-term observations and uncertainties in assessment products limit our ability to evaluate model performance in the Arctic, thereby increasing the uncertainty in future climate projections. In this study, we assess Arctic surface air temperature in our climate model (IPSL-CM6A-LR) against in situ measurements, remote sensing products, and atmospheric reanalyses. Our analysis reveals a persistent winter warm bias over sea ice in IPSL-CM6A-LR ranging from +0.4° to +3.1°C depending on whether the model is atmosphere only or fully coupled with sea ice and ocean. To understand the origin of the biases, we analyze the energy budget across the Arctic Ocean, atmosphere, and sea ice. Our findings show that the model overestimates lateral energy exchanges with midlatitudes, primarily due to excessive northward atmospheric energy transport. This study emphasizes the need to accurately identify and understand model biases to improve future Arctic climate projections.